Dear all,

I encountered a problem when training a recaser. When launching the command

./mosesdecoder/scripts/recaser/train-recaser.perl --first-step 3 --dir
model --corpus corpus.en  --train-script
./mosesdecoder/scripts/training/train-model.perl

the phrase-table ends up having several seemingly identical translation
options:

naţională ||| Naţională ||| 1 ||| 0-0 ||| 30 30 ||| 30 |||
naţională ||| Naţională ||| 1 ||| 0-0 ||| 36 36 ||| 36 |||
naţională ||| Naţională ||| 1 ||| 0-0 ||| 39 39 ||| 39 |||
naţională ||| Naţională ||| 1 ||| 0-0 ||| 4 4 ||| 4 |||

and a segmentation fault occurs when compressing to compact table using the
processPhraseTableMin executable.

Could that be due to a missing encoding normalization step somewhere in the
procedure?
Using a previous version of Moses, the same command above yields just the
line

naţională ||| Naţională ||| 1 1 1 1 ||| 0-0 ||| 109 109 109 ||| |||


Thanks,

Vito Mandorino
-- 
*M**. Vito MANDORINO -- Chief Scientist*


[image: Description : Description : lingua_custodia_final full logo]

 *The Translation Trustee*

*1, Place Charles de Gaulle, **78180 Montigny-le-Bretonneux*

*Tel : +33 1 30 44 04 23   Mobile : +33 6 84 65 68 89*

*Email :*  *[email protected]
<[email protected]>*

*Website :*
*www.linguacustodia.finance <http://www.linguacustodia.com/>*
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support

Reply via email to